Cancer Immune Therapy Edited by G. Stuhler and P. Walden ...

220 10 The Immune System in Cancer: If It Isn't Broken, Can We Fix It?
lease within an environment that is highly immunostimulatory. Regardless of
whether the fuel that lights this immunostimulatory fire is cytokines, adjuvant or
the release of stressful markers of death (Fig. 10.5B), the important result is that
APCs are recruited to the site of tumor antigen release. The APCs can then take up
the released antigens, process them and present them, and, because of the immunostimulatory
environment, will mature and express the full complement of co-stimulatory
molecules [12, 78±81]. In this way, previously immunologically silent antigens
are now `visible' in the context of the complete repertoire of fully activated antigen-presentation
pathways.
Once antigen has been provided in the context of co-stimulatory cross-presentation
by host APCs [78±81], the success of the next phase of this generic tumor vaccination
protocol (Fig. 10.5B) depends upon whether any T cells exist that can respond to epitopes
from TAAs. As discussed above, we know that such T cells exist [5, 82] and how
they might have escaped thymic selection (see Appendix) (Box 10.1). The next challenge,
assuming that the full range of tumor-associated antigens is now displayed on
mature APCs [78±81], is presenting the associated epitopes to the T cell repertoire in
the lymph nodes to where activated APC traffick following acquisition of antigen
and subsequent maturation (Fig. 10.5B). This is precisely what is likely to occur
when the triggers for autoimmune diseases lead to presentation of self antigen to
host T cells (Fig. 10.5A). It is not so surprising, then, and is even encouraging, that
some effective tumor vaccines show signs of autoimmune side effects, in both model
and clinical systems. The best-documented example of this is the induction of vitiligo
in both animals and in human patients following immunotherapeutic treatments
for malignant melanoma. In these cases, activation of immune responses
against melanoma-associated antigens have also led to the induction of reactivity
against similar, or the same, antigens expressed on normal melanocytes [14, 39, 50,
83±86].
Clearly, however, there are serious risks associated with inducing autoimmunity
against self or near-self tumor antigens [76]. Immunostimulatory tumor cell killing
will release not only the relevant tumor antigens, but also a flood of cellular antigens.
All are equally likely to be taken up by suitably activated APCs to be presented
in the context of a full co-stimulatory phenotype. If we postulate that some
low-affinity T cells will exist that recognize tumor antigen-associated epitopes, it is
probable that similar cells also exist to recognize cellular antigen-associated epitopes
with no relevance to the tumor. How can we ensure that the immune system
sifts out those antigens to which we do not want it to react (those with cellular
autoimmune potential) but does develop immunity to those that we do (tumor associated
autoimmunity)? This problem is not specific to induction of antitumor immunity
(Fig. 10.5B); it is also relevant to normal infection, when the death of infected
cells liberates both foreign and normal cellular antigens (Fig. 10.4A) [6]. Recently
it has been suggested that a population of immature DCs [87] continually induces
peripheral (as opposed to central or thymic) tolerance to self antigens, by ingesting
antigens in apoptotic bodies released by normal turnover of tissues [6, 61]
and subsequently presenting them to T cells in the lymph nodes. Since these antigens
are taken up in a tolerogenic manner by the DCs (i.e. the absence of co-sti-